NexusFi: Find Your Edge


Home Menu

 





Taking a Trading System Live


Discussion in Trading Journals

Updated
      Top Posters
    1. looks_one kevinkdog with 260 posts (717 thanks)
    2. looks_two Big Mike with 61 posts (97 thanks)
    3. looks_3 deaddog with 25 posts (24 thanks)
    4. looks_4 swz168 with 20 posts (40 thanks)
      Best Posters
    1. looks_one kevinkdog with 2.8 thanks per post
    2. looks_two swz168 with 2 thanks per post
    3. looks_3 rk142 with 2 thanks per post
    4. looks_4 Big Mike with 1.6 thanks per post
    1. trending_up 96,261 views
    2. thumb_up 1,131 thanks given
    3. group 99 followers
    1. forum 529 posts
    2. attach_file 189 attachments




 
Search this Thread

Taking a Trading System Live

  #241 (permalink)
 
Big Mike's Avatar
 Big Mike 
Manta, Ecuador
Site Administrator
Developer
Swing Trader
 
Experience: Advanced
Platform: Custom solution
Broker: IBKR
Trading: Stocks & Futures
Frequency: Every few days
Duration: Weeks
Posts: 50,396 since Jun 2009
Thanks Given: 33,172
Thanks Received: 101,536


kevinkdog View Post
I'm curious, what were you expecting, and why? It would be interesting to hear your take on things.


I can say that using more than the max drawdown as quit criteria was driven in part by the small size of the max historical drawdown. I definitely do not always use max drawdown as my criteria.

I am afraid to comment more because I really don't have experience with systems that take this few trades.

But I do have experience with systems that take far, far more trades. In those examples, say the system trades 50 times a month, then I am more comfortable because I can look at the trade frequency, duration, win/loss percentage, drawdown, MAE, MFE, or any other slew of metrics and only need to wait a month for meaningful results (50 samples).

I always use drawdown as a quitting point, but never exclusively. The system may be within normal behavior on drawdown by wildly deviating on other metrics, and that would cause me great concern.

I was just surprised you listed drawdown as the sole qualifier. In essence by saying your only quitting point is based on drawdown, you are measuring the entire system based on this one thing. Am I wrong?

Mike

We're here to help: just ask the community or contact our Help Desk

Quick Links: Change your Username or Register as a Vendor
Searching for trading reviews? Review this list
Lifetime Elite Membership: Sign-up for only $149 USD
Exclusive money saving offers from our Site Sponsors: Browse Offers
Report problems with the site: Using the NexusFi changelog thread
Follow me on Twitter Visit my NexusFi Trade Journal Reply With Quote

Can you help answer these questions
from other members on NexusFi?
NT7 Indicator Script Troubleshooting - Camarilla Pivots
NinjaTrader
Deepmoney LLM
Elite Quantitative GenAI/LLM
Exit Strategy
NinjaTrader
ZombieSqueeze
Platforms and Indicators
New Micros: Ultra 10-Year & Ultra T-Bond -- Live Now
Treasury Notes and Bonds
 
Best Threads (Most Thanked)
in the last 7 days on NexusFi
Get funded firms 2023/2024 - Any recommendations or word …
61 thanks
Funded Trader platforms
39 thanks
NexusFi site changelog and issues/problem reporting
26 thanks
The Program
18 thanks
GFIs1 1 DAX trade per day journal
18 thanks
  #242 (permalink)
 kevinkdog   is a Vendor
 
Posts: 3,645 since Jul 2012
Thanks Given: 1,890
Thanks Received: 7,338


Big Mike View Post
I am afraid to comment more because I really don't have experience with systems that take this few trades.

But I do have experience with systems that take far, far more trades. In those examples, say the system trades 50 times a month, then I am more comfortable because I can look at the trade frequency, duration, win/loss percentage, drawdown, MAE, MFE, or any other slew of metrics and only need to wait a month for meaningful results (50 samples).

I always use drawdown as a quitting point, but never exclusively. The system may be within normal behavior on drawdown by wildly deviating on other metrics, and that would cause me great concern.

I was just surprised you listed drawdown as the sole qualifier. In essence by saying your only quitting point is based on drawdown, you are measuring the entire system based on this one thing. Am I wrong?

Mike


OK, I think I understand what you are getting at, and it is something I really haven't discussed. What happens when the system performance isn't bad enough to hit your quitting point, and it is not great either, but is somewhere in between? Maybe it is making money, or breaking even, and it is within the bounds of what Monte Carlo sim says is possible. When do you quit, or otherwise cease trading the system?

My general philosophy is to watch the downside, and let the upside take care of itself. So, in this case, I watch the max drawdown, and as long as that is not hit, let the system perform. I do this because I never know month to month or year to year what particular systems I am trading will do good, which will do bad, and which will just tread water. So, normally I'll just let system be, and not turn them off or on.

But, a couple times a year I rebalance systems I am trading - add in new ones, and cull the underperformers and possibly adjust the position sizing. If capital becomes an issue, I might very well stop trading and swap out a mildly profitable, but underperforming, system for a system that I feel has more potential. The analysis details are never the same, and I don't have strict rules on this. I might, for example, stop trading a system because I no longer like it for some reason - maybe it just doesn't fit me anymore.

I realize that I am sort of talking out both sides of my mouth here. On one side I say "maximum drawdown is my only quitting criteria." On the other side I say "unless I come up with another legitimate reason to cease trading it." I rationalize this by saying the maximum drawdown is a hard, solid, worst case criteria, and will not be violated. At the same time, though, there may be other circumstances which arise which cause the system to fall out of favor. These circumstances might cause me to quit earlier. That is a big gray area.

I can say my plan right now only includes maximum drawdown as sole criteria for quitting. With the small size and account size I am trading this system with, I can't foresee needing the capital for a better system. But, you never know...


I hope this answers your question. Putting this down in words makes me realize there are always caveats to hard and fast rules.

Follow me on Twitter Started this thread Reply With Quote
Thanked by:
  #243 (permalink)
 rk142 
Atlanta, GA
 
Experience: Intermediate
Platform: Ninjatrader
Trading: N/A
Posts: 260 since Dec 2011
Thanks Given: 117
Thanks Received: 328



kevinkdog View Post
I'm actually a little scared someone understands me that well!!!

Just remember, I am just presenting the way I do things, and that is not the only way, and very possibly not even the best way...

Of course! I'm aware of the impulse to make an idol or a fetish out of someone else's approach, and I'm sure in the end my decisions will be different than yours (they already are in some respects). But what you've shared so far resonates with me, and I am picking up things. I haven't changed my process at all yet - I monitor in the same way - but I am thinking about how to incorporate some of what I learned when bringing the next ruleset online.


Big Mike View Post
I always use drawdown as a quitting point, but never exclusively. The system may be within normal behavior on drawdown by wildly deviating on other metrics, and that would cause me great concern.

I don't have Mike or Kevin's experience, so all of Kevin's disclaimers about not knowing the one "true way" apply to me but to the nth power, but I intuitively gravitated on my own towards a sole focus on drawdown. Drawdown is the first thing I look at with a new ruleset. If I establish what the drawdowns look like for a system and I have enough capital to size so that the max drawdown is within my pain tolerance, then I can trade that system. When the drawdown exceeds my tolerance I'm done. If the system has a long term positive expectancy, then pretty much nothing else matters (to me).

I have a tendency to overthink, so if I brought in other metrics I'd probably enable my risk-averse self-saboteur to stop the pain early. It's the system-trade version of tightening your stop on a discretionary trade too early:

Put your system stop where you know you are wrong, make sure that that stop subjects you to an amount of psychological and financial pain that leaves you in a position to fight another day, and let the "trade" play out.

Other, more experienced people (Mike and Kevin) can surely assess a system's performance and quit early, but my fiddling, both with rule-based and discretionary trades, has caused demonstrable deterioration in my results.

All right, I'm a hobby trader, and I'm violating my all important time-management rules by doing trade related stuff before 10AM, so I've got to get out of here....

-RK

Reply With Quote
Thanked by:
  #244 (permalink)
 kevinkdog   is a Vendor
 
Posts: 3,645 since Jul 2012
Thanks Given: 1,890
Thanks Received: 7,338


rk142 View Post
Kevin,

Excellent thread. Thank you.

In response to a question about your approach to optimization you wrote the following:



Could you say a bit more about the tools / decision making process you use to make decisions about the IN/OUT periods you use?

I'm afraid of optimization.

thanks,
RK

One common mistake during walkforward analysis is to surreptitiously optimize the IN and OUT periods. Say, for example, that you run the walkforward analysis with 4 year In period, and 1 year Out period. Walkforward results for that case are good, but not great, so you think "maybe I should use 4 years In, with 2 years out." That case is 200% better, and meets all your goals, so you decide "that's the combination to use. Let's go!"

Stop.

Do you realize what just happened? As soon as you selected a second set of In/Out parameters, reran the results, and selected the best case, you just optimized. Sure, it is not a full optimization, since you only compared two cases, but it was optimization nevertheless. Remembering the rule that optimized results can't be trusted, you have a dilemma here: accept the first run (4 year / 1 year), and then discard the strategy because it did not meet your goals, or accept the second run, and pretend you never optimized.

Once again, I'll admit to doing the above on occasion, although I can't recall it ever ending well. The big question in all this is "is there a way to test multiple IN/OUT periods, and select the best one, while still maintaining walkforward integrity?" The answer, thankfully, is yes. The way to do it is to create, in essence, a second walkforward analysis inside of the first. The way to do this is to run the walkforward analysis, as usual, but leave the last few years of data untouched. I typically will leave three years untouched. Then, with the walkforward data I have, I select the best IN/OUT pair I have, and then run it on the last three years of data. If it passes, then I go on to the next step. If it doesn't, I discard the strategy. But, in either case, at least I have made some effort to select the best IN/OUT combination. The downside to this approach is that you have optimized, and the more optimization you do, the worse off you generally are.

This process would look like this:

1. Years 2000-2008 >> run walkfoward analysis for different combinations of In/Out periods, select the best In/Out

2. Years 2009-present >> run walkforward analysis, using best In/Out determined from Step 1

3a. If walkforward results from 2009-present look good, continue with development

3b. If results do not look good, it is probably best to abandon strategy, rather than try again with another In/Out pair

Follow me on Twitter Started this thread Reply With Quote
Thanked by:
  #245 (permalink)
 
Jura's Avatar
 Jura   is a Vendor
 
Posts: 775 since Apr 2010
Thanks Given: 2,352
Thanks Received: 690


kevinkdog View Post
Right now I am not doing anything with TopStep - the system I am currently running is automated through Tradestation, and traded through their brokerage.

When I was doing the Combine, I was still using the same strategies, but I was manually entereing orders in the T4 platform, which connected to TopStep.

I hope this explains it - if not, please feel free to ask more questions.

What is your take on the use/value of TST for algorithmic automated traders, Kevin?

I know from your earlier posts that your performance was suffering due to needing to manually enter the orders. But let's assume for a moment that one day in the future TST offers support for automated trading. Would you then do another combine or not? And why?

I on the one hand think that doing a successful combine might be a good way to trade a larger capital and make use of the scalability of a strategy. On the other hand, the parameters that TST use are pretty strict and these probably will add to (even) more overfitting during the strategy design process. Plus, designing a profitable strategy is hard enough already without those 'artificial' parameters, I think.

Reply With Quote
  #246 (permalink)
 kevinkdog   is a Vendor
 
Posts: 3,645 since Jul 2012
Thanks Given: 1,890
Thanks Received: 7,338


Jura View Post
What is your take on the use/value of TST for algorithmic automated traders, Kevin?

I know from your earlier posts that your performance was suffering due to needing to manually enter the orders. But let's assume for a moment that one day in the future TST offers support for automated trading. Would you then do another combine or not? And why?

I on the one hand think that doing a successful combine might be a good way to trade a larger capital and make use of the scalability of a strategy. On the other hand, the parameters that TST use are pretty strict and these probably will add to (even) more overfitting during the strategy design process. Plus, designing a profitable strategy is hard enough already without those 'artificial' parameters, I think.


I think once TST gets Ninja working, automated strategies will be viable. I think they got rid of many of the performance metrics.

Creating successful intraday strategies is tough, though. I find it much easier to create swing strategies, lasting days to weeks.

Follow me on Twitter Started this thread Reply With Quote
Thanked by:
  #247 (permalink)
 indextrader7 
Birmingham, AL
 
Posts: 1,065 since Apr 2012


kevinkdog View Post
Creating successful intraday strategies is tough, though. I find it much easier to create swing strategies, lasting days to weeks.

Hey Kevin,

Why do you think this is? I'd be curious to know your thoughts on it.

Reply With Quote
  #248 (permalink)
 kevinkdog   is a Vendor
 
Posts: 3,645 since Jul 2012
Thanks Given: 1,890
Thanks Received: 7,338


indextrader7 View Post
Hey Kevin,

Why do you think this is? I'd be curious to know your thoughts on it.

Great question, Merritt. Here are my thoughts (I encourage people to disagree with my points, and create discussion from which we all can benefit).


Whenever I begin looking at a new strategy, I almost always see if it can be made into a day trading strategy. I define "day trading" as in and out of a trade, or multiple trades, in a single session. There are some nice benefits to such a strategy:

1. No overnight risk from unexpected events, since you are flat

2. Reduced margin requirements, making it easier to trade with large size (although most people should not be doing this)

3. A "working job" type feeling. You fire up the computer in the morning, trade a while, make your daily nut, turn off the computer, go home and play with your kids the rest of the night. Very satisfying way to live.


So, usually when I start development, I select short timeframe bars (1 minute to 5 minute), throw in the "set exit on close" statement to exit at the end of the day, and jump into development.

Nine times out of ten, though, the strategy fails. Regardless of the entry idea (trend, counter trend, whatever) and exit scheme (fixed stops, moving stops, breakeven stops, profit targets, etc.) nothing ever seems to work.

Inevitably, if I like the strategy idea, I'll then open up the timeframe, to 60 minute bars, 240 minute bars, daily bars. I want to see if my idea has any validity at all.

What almost always happens? Performance gets better! Maybe the performance still doesn't meet my goals, but the performance on a daily chart is almost always better than on a 1 minute chart. I've seen it enough times to realize it is more than a coincidence. The question then becomes: why do I see this behavior? Here's what I have come up with:

1. Number of trades and trading costs. Let's say I have daily bar strategy that trades one time per month, or once every 20 bars. That will cost me roughly $25 in trading costs. If I go down to 1 minute bars, the same strategy might trade 10 times per day (once every 120 bars), or $250 in trading costs. That is a huge difference in costs that must be overcome. Add in the fact that 1 minute moves are smaller than daily moves, and it gets even harder.

2. There seems to be more randomness in the data as you go to smaller timeframe bars. Look at a 1 minute chart of ES and most days it is just narrow range noise. It is harder to find the true price path when the random noise level is high. Daily bars, as an alternative, seem to have more trends. Of course, where I see randomness in data could just be due to other biases I have floating around in my brain.

3. Entry and exits become a much more important part of the system, when you have small stops and targets, as most day trading systems are set up to be. So, I must have really great entries and exits, ones with very good edges. But, the better the entry, the harder it is to find during development. Plus, miss the entry by a tick, and you may lose a good percentage of your profit. If you are swing trading with daily bars, a tick or two at entry probably won't mean as much, relative to overall size of an average trade.

4. With tick charts or 1-5 minute charts, think about who you are trading against. Many times it is high frequency trading firms, who probably have better entries than you, and have a speed advantage over you. I feel the impact of the pros is less noticeable at higher timeframes, although I realize many pros trade daily bars, too.

5. With most strategies, as I mentioned before, I find the less trades there are, the better. This could be due to trading costs, but it also could be due to a very bad reason: maybe you think you have an edge, but with few trades, the statistical confidence that you have an edge is a lot lower. Put another way, if I had 2 strategies that averaged $50 profit per trade, and one had 100 trades over the past ten years, and another had 1,000, I'd always pick the 1,000 trade strategy (so would every rationale person). But, the reality is that the 100 trade strategies are a lot easier to find - maybe it is because they aren't really edges at all?

I really wish that all my strategies were day trading type strategies. In actuality, probably 9 out of 10 are the exact opposite. In fact, my best strategy over the past 4 or 5 years holds a position for weeks to months.


I would love to hear the opposite viewpoint to what I have experienced - why someone finds it easier to develop intraday strats.

Follow me on Twitter Started this thread Reply With Quote
  #249 (permalink)
 
Big Mike's Avatar
 Big Mike 
Manta, Ecuador
Site Administrator
Developer
Swing Trader
 
Experience: Advanced
Platform: Custom solution
Broker: IBKR
Trading: Stocks & Futures
Frequency: Every few days
Duration: Weeks
Posts: 50,396 since Jun 2009
Thanks Given: 33,172
Thanks Received: 101,536

Years ago I did exclusively intraday type strategies. I also believe automation should leverage its strengths of speed (more trades, faster reaction times).

However, over the years I've drifted to portfolio strategies that trade a basket of stocks or futures. This gives me the increased frequency I demand and creates some intraday trades as a group.

I find this to literally be the holy grail and cannot encourage it enough over the traditional one strategy one instrument system.

Portfolio trading is the best way to manage risk IMO. After you do that, profits are the easy part.

Sent from my LG Optimus G Pro

We're here to help: just ask the community or contact our Help Desk

Quick Links: Change your Username or Register as a Vendor
Searching for trading reviews? Review this list
Lifetime Elite Membership: Sign-up for only $149 USD
Exclusive money saving offers from our Site Sponsors: Browse Offers
Report problems with the site: Using the NexusFi changelog thread
Follow me on Twitter Visit my NexusFi Trade Journal Reply With Quote
  #250 (permalink)
 Luger 
Nashville, TN
 
Experience: Intermediate
Platform: NinjaTrader
Broker: IB
Trading: NQ ES
Posts: 468 since Feb 2011
Thanks Given: 323
Thanks Received: 543



kevinkdog View Post
Inevitably, if I like the strategy idea, I'll then open up the timeframe, to 60 minute bars, 240 minute bars, daily bars. I want to see if my idea has any validity at all.

What almost always happens? Performance gets better! Maybe the performance still doesn't meet my goals, but the performance on a daily chart is almost always better than on a 1 minute chart. I've seen it enough times to realize it is more than a coincidence.

I would love to hear the opposite viewpoint to what I have experienced - why someone finds it easier to develop intraday strats.

Sorry, I'm not going to disagree. I didn't have success with automated strategies until I took the leap from intraday to swing. Funny thing is that swing is the one time frame that I never thought I would trade, it felt like limbo-land between technicals and fundamentals, but apparently it is a decent place to trade.

As to the reasons, I believe that @Fat Tails could give us a math lesson that would prove that longer time frames should be more profitable. He posted it in another thread or threads because the timeframe topic is usually addressed as a sub-topic to something else. Though I believe it gets down to the points you made about noise and costs.

Though one thing I personally find interesting is that many times I end up building systems on a lower time frame than expected given the trade duration. I don't think most people would build a system with a 3.5 day average trade duration based on a 10-minute chart (about 140 bars per trade). Then again, I have not heard statistics of many profitable systems that people run for any period of time.

Reply With Quote
Thanked by:




Last Updated on January 6, 2016


© 2024 NexusFi™, s.a., All Rights Reserved.
Av Ricardo J. Alfaro, Century Tower, Panama City, Panama, Ph: +507 833-9432 (Panama and Intl), +1 888-312-3001 (USA and Canada)
All information is for educational use only and is not investment advice. There is a substantial risk of loss in trading commodity futures, stocks, options and foreign exchange products. Past performance is not indicative of future results.
About Us - Contact Us - Site Rules, Acceptable Use, and Terms and Conditions - Privacy Policy - Downloads - Top
no new posts